CS Theory at Northeastern

Home Courses Seminar

Computer Science Theory at Northeastern


Click here to subscribe to our mailing list.
Click here to subscribe to our Google Calendar.

Theory Seminar

This semester the theory seminar will be held on Wednesdays 12:40 - 2:10 PM, and is being co-organized by Sushant Agarwal, Rose Silver, and John Wilkins. Write to John with suggestions for speakers!

Spring Semester, 2024

April 17
12:40 PM
Hastings 209
Rachel Redberg
Data-adaptive methods for differential privacy.
Abstract
Differential privacy (DP) provides rigorous guarantees which can be used to provably bound the privacy loss of running an algorithm on sensitive data. But there is still a large gap between theory and practice which presents challenges to the widespread deployment of DP in machine learning (ML) applications. I plan to talk about several tools and algorithms to bridge this gap.

Using a data-dependent version of DP can help to improve the privacy-utility trade-off of a DP algorithm — but this must be done with great care as data-dependent privacy losses are themselves a function of sensitive data. First I will discuss how to privately publish data-dependent privacy losses for the objective perturbation mechanism. I’ll then demonstrate how data-dependent privacy losses can be used to develop DP algorithms which can adapt to favorable properties of the data, in order to achieve a better privacy-utility trade-off. I’ll conclude by returning to the objective perturbation mechanism, and discuss new tools and privacy analyses that allow it to compete with more modern algorithms.

Speaker Bio: Rachel Redberg is a postdoctoral fellow at Northeastern’s Institute for Experiential AI. She obtained a PhD (computer science) in December 2023 from UC Santa Barbara and a BA (applied math) from UC Berkeley in 2015.
April 10
12:40 PM
Hastings 209
Yair Zick
A Simple Vision for Fair Division
Abstract
Mar 27
12:40 PM
Hastings 209
Connor Wagaman
Time-Aware Projections: Truly Node-Private Graph Statistics under Continual Observation
Abstract
Mar 20
12:40 PM
Hastings 209
Sammy Khalife
On the expressive power of Graph Neural Networks and the role of the activation function
Abstract
Mar 13
12:40 PM
Hastings 209
Mahdi Haghifam
Information Complexity of Stochastic Convex Optimization: Applications to Generalization and Memorization
Abstract
Feb 28
12:40 PM
Hastings 209
Ilya Volkovich
Synergy between Circuit Obfuscation and Circuit Minimization
Abstract
Feb 21
12:40 PM
Hastings 209
Evi Micha
Fair and Efficient Social Decision Making
Abstract
Feb 14
12:40 PM
Hastings 209
Rebecca Lin
Graph Threading
Abstract
Feb 7
12:40 PM
Hastings 209
Neha Makhija
Using ILPs and Flow Algorithms as a Unified Approach for (Reverse) Data Management Problems
Abstract
Jan 17
12:40 PM
Hastings 209
Antonis Skarlatos
Dynamic algorithms for k-center on graphs
Abstract
Jan 10
12:30 PM
Hastings 204
Jessie Finocchiaro
Theoretical ML
Abstract

Fall Semester, 2023

Spring Semester, 2023

Fall Semester, 2022

Spring Semester, 2022

Fall Semester, 2021

Spring Semester, 2020

Fall Semester, 2019

Spring Semester, 2019

Fall Semester, 2018

Spring Semester, 2018

Fall Semester, 2017

Spring Semester, 2017

Fall Semester, 2016

Spring Semester, 2016

Fall Semester, 2015

Prehistorical Theory Seminars